A reliable automation isn't one that never fails; it's one that preserves the work, makes failure visible and enables safe recovery.
Persist before delivering
Record the event and destination before calling a vendor. User confirmation shouldn't depend on CRM or e-mail availability.
Design for redelivery
Idempotency keys, jittered backoff and retry limits prevent duplication and call storms.
Give the exception an owner
Dead letters, alerts, SLOs and audited replay turn technical failure into a treatable operational process.